|
The implicit nature of the data
warehouse to undergo changes makes its maintenance very crucial.
Post-implementation maintenance issues can prove to be critical to
success of data warehouse projects. In addition to issues that are
operational in nature, data warehouse maintenance includes issues
related to growth and upgrades of the data warehouse itself.
The flip-side of the coin is the
complexity involved in data warehouse maintenance. Therefore, a
structured approach to maintenance is necessary to address maintenance
issues associated with data warehouses. Information Architects data
warehouse maintenance methodology is a structured scalable approach that
ensures smooth operability of the warehouse and is geared to provide
flexible enhancements to warehouse architecture.
Information Architects data warehouse
maintenance methodology covers the entire spectrum of maintenance
categories to provide comprehensive maintenance solutions to clients in
pre- and post-implementation scenarios.
The maintenance activities have been
divided into four categories:
·
Corrective Maintenance
Corrective maintenance consists of two distinct processes:
Defect
Identification results from careful monitoring of various
data warehouse processes. This process horizontally caters the
data warehouse; from ETL batch monitoring to OLAP reporting.
This process utilizes skilled and experienced human resources
and automated tools.
Defect
Rectification consists of “Triage”, “Investigation” and
“Fix” phases to rectify defects identified by the earlier
process.
Example
scenarios that qualify for this category of maintenance include
processing partial changes in sources systems as updates to the
data warehouse and impact analysis of business changes in view
of historical data in the data warehouse.
·
Adaptive Maintenance
The framework
ensures easy transition of warehouse if there is a change in the
operating environment of the warehouse such that the transition
is transparent to the users. This change may be in the hardware
infrastructure, operating system or the DBMS.
Extended
scenarios that also qualify for this category of maintenance
include impact of change in the underlying infrastructure on the
structure of data warehouse itself and analysis of such changes
on end-user queries and reports.
·
Preventive Maintenance
In parallel to
the reactive nature of “Corrective Maintenance”, the framework
provides a proactive approach of “Preventive Maintenance”. The
maintenance team continuously reviews the data warehouse to
point out potential defects even if no prior incidents are
logged.
Example
scenarios that qualify for this category of maintenance include
refinement of aggregate and summarization structures and
conduction of security audits.
·
Perfective Maintenance
The framework
can be leveraged for rapid enhancement and refinement of the
warehouse to provide improved functionality and better user
experience.
Example
scenarios that qualify for this category of maintenance include
performance tuning, query tweaking, qualifying a part of
historical data to be purged, identification of new aggregate
and summarization structures and identifying “holes” in the
data. Furthermore, usability audits for data warehouse users may
be conducted to identify scenarios that may lead to discovery of
useful business insight.
Maintenance Priority
Allocation
Taking a bottom-up
approach, DBMS related issues take highest priority and call for
urgent response; followed by ETL and then OLAP. However, certain
issues may be ranked higher in priority depending upon client’s
requirements.
Information Architects' team specializes in providing comprehensive
data warehouse maintenance service to clients . There is a dedicated
team of experts to handle the entire spectrum of data maintenance issues
24/7/365.
Information Architects employ an
open architecture (vendor independent), open tool (product agnostic)
policy. We help clients develop solutions in their own preferred
environment configured to maximally leverage existing resources,
therefore providing flexibility to accommodate variable budgets, and we
are equally well versed with Oracle, SQL Server, Teradata and Sybase. |